This study investigates unsupervised machine learning (ML) for anomaly detection in solar photovoltaic (PV) power generation data from 2019 to 2023. An unsupervised approach is selected to overcome the absence of pre-labeled fault data, enabling the autonomous identification of operational patterns. Following data preparation, K-means clustering (k=3) identifies distinct operational patterns, specifically characterizing regimes such as optimal performance (Cluster 2) and low energy output attributed to adverse weather conditions (Cluster 1). These clusters are subsequently visualized using principal component analysis (PCA) to validate their distinct separation. An isolation forest model is then employed for anomaly detection, identifying 17 significant deviations. These anomalies occur most frequently in 2020, coinciding with the COVID-19 pandemic period. Many fall outside the typical energy range of 2.0–3.2 kWh/day and are associated with non-ideal weather conditions. This finding demonstrates that unsupervised ML provides a scalable framework for monitoring PV system health, enhancing reliability, and supporting preventive strategies.
Although natural light is the standard reference for color rendering evaluation, limited efforts have been made to develop a lighting system that achieves high color rendering while replicating the spectral characteristics of natural light. This paper proposes a method for developing LED lighting that mimics the spectral ratio characteristics of natural light and achieves high color rendering. Based on the analysis of measured natural light spectra, spectral ratio characteristics are derived according to changes in color temperature. Simulations are conducted by adding light sources with specific peak wavelengths (405 nm, 630 nm) to commercial LED lighting. Based on the simulation results, a development approach for natural light LED lighting with high color rendering is proposed. Experimental results demonstrate that a color rendering index (CRI) of 90 or higher (including R9 and R12 ≥ 80) can be achieved for lighting systems with baseline CRI values of 80, 85, and 90.
Illegal nighttime fishing remains difficult to monitor because vessels often deactivate the Automatic Identification System (AIS). This study presents a supervised deep neural network (DNN) approach for detecting nighttime fishing vessel lights using the Day/Night Band (DNB) data from the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard the Suomi NPP and NOAA-20 satellites. The model integrates DNB radiance features with lunar illumination information to reduce false detections caused by moonlight. The dataset comprises summer-season observations (2020–2021) in waters near Jeju Island, South Korea, with labels derived from temporally matched AIS records. The proposed DNN is evaluated using a stratified train–test split and compared with conventional machine-learning baselines. Experimental results demonstrate improved performance, achieving an F1 score above 0.90, indicating the robust detection capability under low-light maritime conditions. These findings highlight the potential of VIIRS DNB data combined with deep learning for large-scale nighttime maritime monitoring beyond AIS-dependent systems.
This research aims to enhance the physical-layer security (PLS) of multi-input multi-output orthogonal frequency division multiplexing (MIMO-OFDM) wireless communication systems. The proposed method integrates Rivest–Shamir–Adleman (RSA)-based subcarrier index scrambling with secure hash algorithm (SHA)-256 hashing to ensure both confidentiality and integrity without altering the transmitted waveform. The scheme is implemented on a 2×2 Universal Software Radio Peripheral (USRP)-based MIMO-OFDM software-defined radio (SDR) testbed and evaluated using symbol distribution analysis, bit-error-rate (BER) performance, and security assessment. Experimental results indicate that RSA-based scrambling substantially increases symbol randomness while preserving Shannon-consistent BER performance across multiple modulation orders, with minimal penalty in signal-to-noise ratio (SNR). Additionally, security analysis demonstrates that the combination of RSA-2048 and SHA-256 is resistant to Wiener’s attack and the common modulus attack, thereby supporting efficient and secure real-time MIMO-OFDM communication.
This study aims to analyze the influence of inter-electrode spacing on grounding resistance in high-voltage transmission networks. A simplified analytical model is applied to a case study on the Djiri-Ngo 220 kV line (Republic of the Congo), considering two representative soil types: clayey and siliceous sand. The grounding resistance is calculated by varying the number of electrodes and their spacing. The results show that increasing electrode spacing reduces grounding resistance. In certain configurations, the improvement exceeding 50 % when the spacing is increased from 5 m to 25 m. A saturation threshold is identified, beyond which further increases in spacing yields diminishing returns. Electrode spacing proves to be a key design factor, sometimes more influential than the number of electrodes. The proposed parametric geometric analysis offers a practical and cost-effective strategy for grounding system design, emphasizing the importance of adapting configurations to local geotechnical conditions.
